Addressing remitting behavior using an ordinal classification approach

Abstract

Remittance flows have drawn the attention of international development community interested in enhancing their potential benefits in the recipient communities. This papers deals with the migrants’ remitting patterns, addressing this economic behavior by a classification approach rather than the traditional regression one. Five nominal and two ordinal classifiers were compared in order to verify the nature of the problem and to obtain a model which predicts the remittance levels sent by migrants according to their individual characteristics. The best performance was achieved by the support vector machine with ordered partitions, an ordinal classifier based on binary decomposition, and thus three remitting profiles for immigrants were drawn from the support vectors obtained. As result, the proposed model can be used as a tool for better factoring remittances flows into the design of policies and programs in the migrants’ home country.